The Rev. T. R. Robinson's Description of an improved Anemometer. 1 63 



to those of the actual anemometer, are fixed on an arm in which, by means of 

 a long slit, the axis of rotation can be shifted to any position ; this axis causes a 

 graduated circle to measure 0, the zero of which is determined by a vane above. 

 The axis is shifted till the two pressures are equal, when, of course, a and a' 

 are inversely as its distances from the two centres. In reducing this to prac- 

 tice, however, I found a difficulty which I had not anticipated. Since the forces 

 on the hemispheres are as F^ sin^ 6, 1 concluded they would be at the maximum 

 at 90", and vanish at 0° or 180''; and began by observing them in the first of 

 these positions. To my great surprise, I found that the equilibrium there is un- 

 stable, so that if the angle be changed the least either way, the concave predo- 

 minates. This makes it hard to ascertain the true point of balance, as the 

 direction of the wind is ever changing ; but nevertheless I think I am war- 

 ranted in concluding, with some confidence, from sixteen experiments made in 

 four days with winds from a moderate breeze to a hard gale, 



^ = 4-01]; 

 a 



or, in round numbers, the action on the concave is four times that on the 

 convex. i 



I was the more surprised at this predominance of the concave when the 

 arm is inclined to the wind, because then the part HK of its convex acts 

 against it. 



From some other angles I obtained, though by fewer observations. 



Beyond this it is impossible to go, as there the convex surface apparently ceases 

 to act. In fact, on removing the hemisphere DK, the other remains as in the 

 wood-cut, making 6 — 210' nearly, and oscillating as the tlirection of the wind 

 changes. Whether this arises from the minus-pressure at the segment lA, or 

 from the wind which passes at B eddying into the concave, I cannot decide ; 



VOL. XXII. z 



